arXiv subjects
Publications and source records attributed to Yang Hu.
We explore the classical question of enumerating the homotopy types of gauge groups of principal $\mathrm{SU}(n)$-bundles over $S^{2m+1}$ in the unstable range $m\geq n$. Using computations of certain $p$-local Samelson products, we obtain lower bounds on the number of such homotopy types. We also present a discussion of the last stable case.
We show how to implement a dynamic array $A[1, n]$ with symbols from a fixed alphabet $Σ$, while supporting $O(1)$-time queries and updates, and using a total space of $$ \log \binom{|Σ|}{m} + \left(1 + O\left(\frac{\log \log n}{\log n}\right)\right) \cdot \left(\sum_{σ\in Σ} f_σ\log (n / f_σ)\right) + n / \text{polylog } n $$ bits, where $f_σ$ denotes the frequency of each symbol $σ\in Σ$ and $m$ denotes the number of distinct symbols with non-zero frequencies. This resolves a long-standing open question as to whether one can achieve space bounds close to that of arithmetic coding, while supporting $O(1)$-time operations, whenever the entropy is at least $n/\text{polylog } n$. We also prove a nearly matching space lower bound: up to a factor of $O(\log \log n)$, the entropy-dependent multiplicative overhead of our construction is optimal among $O(1)$-time solutions when $|Σ|=O(\sqrt n)$ and the entropy $\sum_{σ\in Σ} f_σ\log (n / f_σ)$ lies between $n/\log^{O(1)}n$ and $(1/100)n\log n$. Finally, we present several applications of our results, resolving two open problems having to do with space-efficient dictionaries and filters.
The dynamic dictionary is a fundamental data structure that maintains a set $S\subset [U]$ of size $n$ (we assume $n=U^{1-Θ(1)}$), supporting insertions, deletions and membership queries. Previous works mostly focused on constructing dictionaries that support operations in $O(1)$ time and use space as close to the \emph{information-theoretic bound} of $\log\binom{U}{n}$ bits as possible. In this paper, we study \emph{difference-encoded} dictionaries, which are dictionaries that use space close to the gap entropy $\text{gap}(S):=\sum_{i=2}^{|S|}\left(\lceil\log(x_i-x_{i-1}+1)\rceil+1\right)$ bits to store the set $S=\{x_1<\dots<x_n\}$. On many real-world datasets where the keys are clustered, we have $\text{gap}(S)\ll \log\binom{U}{n}$, making difference-encoded dictionaries more favorable than standard dictionaries in practice. Prior to this work, the best dynamic difference-encoded dictionary is by Blandford and Blelloch [SODA'04], whose construction supports operations in $O(\log n)$ time and uses $O(\text{gap}(S))$ bits of space. In the static case, Gupta, Hon, Shah and Vitter [DCC'06] presented a dictionary that uses $$ \text{gap}(S)+O(n\log\log U) $$ bits of space and supports membership queries in $O(\log\log n)$ time. In this work, we go beyond these bounds and fully settle the optimal time-space tradeoff for difference-encoded dictionaries. For an arbitrary parameter $0<\varepsilon<1/4$, we construct a dynamic dictionary that supports operations in $O(\log\varepsilon^{-1}/\log\log\varepsilon^{-1})$ expected amortized time and uses $$ \text{gap}(S)\cdot(1+O(\varepsilon))+O\left(n\log\frac{\text{gap}(S)}{n}\right) $$ bits of space. We also prove a matching lower bound, showing that our time-space tradeoff is optimal even in the \emph{static} case.
Modern pre-silicon design-space exploration (DSE) follows a coarse-to-fine workflow: low-cost evaluators screen candidate spaces, while detailed evaluation is reserved for a shortlist. Wafer-scale systems strain both stages. Architecture choices induce coupled physical implications for reticle compliance, wafer tiling, die area, D2D capability, boundary access, and placement, so the design space cannot be treated as an unconstrained Cartesian product. Meanwhile, detailed evaluation is too expensive to cover the resulting space, whereas analytical-to-reference ranking inversions make a fixed shortlist unreliable. We present Fovea, a reusable methodology for workload-specific wafer architecture selection rather than a fixed wafer template. Fovea first performs physical-implication-aware design-space formulation to construct a distinct modeled-feasible space while preserving cross-dimensional trade-offs and applying only evaluator-preserving local reductions. It then performs Decision-Domain-guided cross-fidelity refinement. Paired in-domain calibration estimates workload- and space-specific analytical-to-reference disagreement, which parameterizes reference-consistent performance intervals and induces a Decision Domain for selective designated-reference evaluation. Under a valid domain-wide disagreement bound, this domain contains the designated-reference optimum; the sampling-based implementation is evaluated empirically on exhaustive-reference design spaces. Across ten reference-verifiable design spaces and seven LLM-training workloads, Fovea with 10% paired calibration recovers the exhaustive designated-reference optimum in all 70 evaluated pairs while achieving 4.13x average and 7.80x maximum end-to-end speedup.
Pathology foundation models (PFMs) provide strong tissue representations and have become central to digital pathology. However, deployment in disease-specific settings is limited by 1) the high computational cost of billion-parameter PFMs and 2) distribution mismatch and non-biological bias inherited from pan-cancer, multi-centre pre-training, including site-specific signatures and imbalanced disease prevalence. These factors can encourage shortcut learning and under-emphasise subtle morphology required for reliable modelling of a specific cancer type. We present SmartStu (a Smart Student), a framework to customise compact, breast-cancer-specific PFMs via distillation whilst mitigating confounding. SmartStu distils representations from multiple teacher PFMs into a lightweight student backbone. Crucially, we introduce adversarial distillation that leverages a dedicated noise model trained to predict nuisance, edge-dominated cues on the distillation set. Using this noise model as a counterexample, the adversarial objective encourages the student to recognise, yet suppress, features predictive of nuisance targets. We further incorporate multi-teacher ensemble distillation and an auxiliary self-supervised objective with artefact injection. We validate SmartStu on three external cohorts (Yale HER2, SLN-Breast, and BRACS) with multiple tiny backbones. SmartStu yields breast-cancer-specific PFMs that are over $30\times$ smaller than general PFMs whilst largely preserving, and sometimes improving, downstream performance measured by balanced accuracy (bAcc) and AUC. Code is available at https://github.com/zwchen03/advDistall.
Pathology foundation models (PFMs) provide strong tile-level representations via self-supervised pre-training on large-scale pathology images. Yet, PFMs are developed under diverse and often opaque data, architecture, and objective choices, inducing latent representational biases that limit robustness and obscure what each model specialises in. We present AdaFusion, a lightweight adaptive fusion framework that integrates complementary signals from multiple frozen PFMs through (1) low-dimensional feature compression and (2) a sample-conditioned gating module that reweights model-wise (and optionally channel-wise) contributions. Beyond improving predictive accuracy, AdaFusion provides contribution-driven interpretation that offers evidence consistent with model-specific preferences and synergistic interactions across tissue phenotypes. We evaluate AdaFusion on three public benchmarks spanning treatment response prediction, prostate cancer grading, and spatial gene expression inference. AdaFusion consistently outperforms individual PFMs and other fusion baselines, while providing interpretable tissue visualisation which aligns model preferences with morphological patterns. Code is available at: https://github.com/xyx-98/PathoOracle.
Pathology foundation models (PFMs) provide strong tile-level representations but remain difficult to interpret at the cellular and microenvironmental scales that underpin clinical reasoning. We introduce Graph-Interpreter (G-Interp), a graph-distillation framework that equips a frozen PFM teacher with a cellular-communication-level "plug-in" interpreter, without modifying the teacher. For each tile, we segment cells as graph nodes and construct a microenvironment graph based on spatial adjacency. Graph neural network (GNN) students distil the PFM embedding, whilst learning attention-based message passing that yields node- and edge-level importances. We interpret these importances as cell-cell communication evidence, providing fine-grained explanations of how PFMs encode microenvironmental context. To stabilise distillation when graph abstraction is imperfect, we employ a lightweight auxiliary student to supply complementary visual cues and condition graph message passing, while keeping the primary interpretability signal graph-derived. We evaluate explanation faithfulness by mapping graph-selected evidence back to the image using instance masks and measuring teacher sensitivity under targeted vs non-target occlusions. Across multiple histopathology tasks, G-Interp produces highly scalable, microenvironment-aware explanations, while maintaining competitive predictive performance.
IAM misconfigurations are a major cause of privilege escalation (PE) attacks in the cloud, leading to data breaches and major financial losses. Existing PE detectors have two main limits: they cover only some PE types, so many attacks are missed, and they require full access to cloud configurations, which customers may not want to share because of sensitive information. We present TAC, the first IAM PE detection framework that supports both whitebox and greybox analysis for Amazon Web Services (AWS). To improve coverage, we systematically study how permissions are acquired in AWS IAM and identify five PE categories. All five share one pattern: permissions spread across entities. We define this as permission flows and manually extract 219 templates from more than 14,000 AWS operations. Based on this, we build TAC-WB, a whitebox detector with broad PE coverage. We also build TAC-GB, the first greybox PE detector, which works with partial configurations. Customers can choose which entities to reveal and whether to answer questions about permissions. TAC-GB uses a dynamic query process that adapts to each response and uses reinforcement learning with graph neural networks to ask the most useful questions while reducing interaction. We also create TAC-Bench, a benchmark with 2,500 tasks reflecting real-world IAM misconfigurations. Experiments show that TAC-WB finds all PEs missed by prior tools, while TAC-GB outperforms other greybox methods and often matches whitebox methods even with limited query budgets.
Quadratic probing is one of the most widely used open-addressing hash-table schemes in practice, but after more than half a century, even its most basic performance guarantees remain poorly understood. In this paper, we revisit quadratic probing through the lens of a smoothed variant in which each key follows a random probe sequence where its $k$th probe is expected at offset $Θ(k^2)$. This is simultaneously a toy model for better understanding regular quadratic probing and a natural hashing scheme in its own right. We analyse smoothed quadratic probing for both Robin Hood ordering and anti-Robin Hood ordering and reveal a surprising separation: At load factor $1-\varepsilon$, anti-Robin Hood achieves an expected query time of $Θ(\log \varepsilon^{-1})$, which matches the conjectured expected average successful query time for regular quadratic probing, while Robin Hood falls short at $Θ(\varepsilon^{-1/2})$. Our analysis generalises to degree-$d$ probing for any $d \ge 1$ with expected query time $O(\max(\log \varepsilon^{-1}, \varepsilon^{1-2/d}))$ for anti-Robin Hood and $Θ(\varepsilon^{-1/d})$ for Robin Hood. Finally, we go beyond smoothed analysis: using the probabilistic method, we show that for every $d \ge 2$, almost every random fixed-offset degree-$d$ probing sequence achieves expected query time $O(\log \varepsilon^{-1})$ under anti-Robin Hood ordering, simultaneously over all admissible table sizes and load factors. Thus, while quadratic probing itself remains elusive, we prove that essentially all quadratic-probing-like fixed-offset schemes achieve the ideal performance under the anti-Robin Hood ordering.
Diffusion policies have shown promising empirical performance in representing and learning complex maneuvers for robots using behavior cloning (BC). In this paper, we explore training diffusion policies from scratch using reinforcement learning (RL) for multi-task robotic manipulation. Specifically, we aim to train a single diffusion policy for block-pushing tasks with multiple shapes. The proposed framework features a simple policy loss function, which is a reweighted evidence lower bound used in BC-based diffusion policy training and can seamlessly serve as the policy learning module in RL algorithms. To address the exploration challenges arising from the absence of demonstrations, we incorporate reverse curriculum generation and objective-centric representations. Combined with the expressiveness of diffusion policies, our design supports learning of multi-task block-pushing policies in our sparse-reward simulation setting. We further evaluate whether the trained diffusion policy transfers in zero-shot to real-world tasks under varying environmental conditions including goal positions, block shapes, block weights and surface friction, providing evidence that this pipeline can transfer to our real-world block-pushing setup under the tested variations.
We study the problem of constructing concurrent objects in a setting where $P$ processes run in parallel and interact through a shared memory that is subject to write contention. Our goal is to transform hardware primitives that are subject to write contention into ones that handle contention gracefully. We give contention-resolution algorithms for several basic primitives, and analyze them under a relaxed, roughly-synchronous stochastic scheduler, where processes run at roughly the same rate up to a constant factor with high probability. Specifically, we construct read/write registers and CAS registers that have latency $O(\log P)$ w.h.p. under our scheduler model, using $O(1)$ hardware read/write registers and, in the case of our CAS construction, one hardware CAS register. Our algorithms guarantee performance even when their operations are invoked by an adaptive adversary that is able to see the entire history of operations so far, including their timing and return values. This allows them to be used as building blocks inside larger programs; using this compositionality property, we obtain several other constructions (LL/SC, fetch-and-increment, bounded max registers, and counters). To complement our constructions, we give a trade-off showing that even under a perfectly synchronous schedule and even if each process only executes one operation, any algorithm that implements any of the primitives that we consider, uses space $M$, and has latency at most $L$ with high probability must have expected latency at least $Ω(\log_{ML} P)$.
Moon and Moser proved a sharp edge-extremal theorem for Hamilton cycles in balanced bipartite graphs with minimum degree at least $k$. Li and Ning obtained spectral analogues for Hamiltonicity in balanced bipartite graphs of order $2n$ and for traceability in nearly balanced bipartite graphs with part sizes $n$ and $n-1$, under the assumption $n\ge (k+1)^2$. We show that their sharp spectral thresholds remain valid in the linear ranges $n\ge 2k$ and $n\ge 2k+1$, respectively. More precisely, we determine the extremal values of the adjacency spectral radius and the signless Laplacian spectral radius for non-Hamiltonian balanced bipartite graphs with minimum degree $δ(G)\ge k$, and for non-traceable nearly balanced bipartite graphs with $δ(G)\ge k$. In each case, the extremal graph is unique up to isomorphism. Our proof is based on feasible graph parameters: parameters that increase under edge addition and are nondecreasing under Kelmans operations. This yields Moon--Moser type extremal theorems for a general class of parameters, from which the spectral results follow.
Generalizing sleep staging models to unseen datasets is challenging, and typical domain generalization (DG) methods often rely on multiple source domains or domain labels that are rarely available in practice. We tackle the stricter and more practical setting of single-source domain generalization: training on a single labeled source dataset, without domain labels or access to target data. We present SleepBand, a physiology-guided framework that embeds oscillatory priors via a learnable Morlet filter bank and a structured integration-and-recalibration pipeline. This anchors representations to domain-invariant sleep rhythms (e.g., slow waves, spindles), reducing reliance on dataset-specific artefacts. On five public datasets, SleepBand achieves state-of-the-art SDG performance and remains competitive under leave-one-domain-out (multi-source) DG. Analyses show that the learned filters align with canonical neurophysiology and that robustness stems from focusing on narrowband, physiologically meaningful cues. Our results suggest that principled, physiology-aware inductive biases are a promising path for robust single-domain sleep staging. Code is available at https://github.com/lzcn/sleep-band
Large language models (LLMs) are increasingly used in prefill-only workloads, where end-to-end latency is dominated by the prefill phase. For long-context prefill, communication overhead grows with sequence length and quickly becomes a bottleneck on conventional GPU systems, making wafer-scale chips (WSCs) a promising substrate due to their high communication bandwidth and large aggregate compute and memory capacity. A natural way to accelerate prefill is to partition a long input sequence into multiple chunks and execute them in a finer-grained pipeline across devices. However, directly applying this idea to long-context prefill on WSCs remains challenging. First, causal dependency across chunks causes KV cache to accumulate unevenly across pipeline stages, creating severe memory imbalance and limiting the feasible sequence length. Second, later chunks require more attention computation because each chunk depends on preceding chunks, leading to chunk-level latency imbalance. To address these challenges, we present MOCAP, a memory-orchestrated chunked pipelining framework for prefill-only LLM inference on WSCs. MOCAP introduces Memory-Balanced KV Reallocation (MBKR) to alleviate memory imbalance by redistributing KV cache across pipeline stages, thereby extending the feasible sequence length. It further incorporates Latency-Balanced Chunk Partitioning (LBCP) to balance chunk execution cost under both attention-cost growth and KV reallocation overhead, improving pipeline efficiency. Experimental results show that, compared with GPipe, MOCAP achieves 76.4\% lower end-to-end latency and 3.24$\times$ higher throughput on average. MOCAP also extends the maximum supported sequence length by up to 1.31$\times$ compared with Terapipe.
Latent actions provide a compact interface between action-free video and downstream decision-making, yet existing Latent Action Models (LAMs) force every transition through a fixed-capacity bottleneck. We identify a bottleneck trade-off: overly tight codes can discard transition cues needed for action alignment, while overly loose codes preserve additional transition variation that must be resolved when alignment labels are scarce or narrowly distributed. FlexLAM replaces this fixed capacity with variable-length latent actions trained by nested dropout, yielding prefix-valid codes that capture compact transition structure first and add detail only when needed, without new architectures or losses. A single FlexLAM matches or surpasses separately trained fixed-capacity LAMs at every evaluated token budget under standard scarce-label supervision and under a low-return single-task alignment stress test, indicating that FlexLAM is not merely adjustable at inference time but learns a better latent-action interface at the same token budgets. The same model supports inference-time token-budget adjustment without retraining, and FlexLAM improves Ego4D transition reconstruction. These results suggest that variable-length latent actions are an architecture-free, drop-in upgrade to the fixed-capacity bottleneck in latent action models, latent-action world models, and video-pretrained action interfaces.
With the rapid advancement of unmanned aerial vehicle (UAV), ensuring the security of communication links among UAVs has become crucial. In this paper, we propose a novel physical layer key generation scheme based on channel codeword reconstruction. In UAV communications, the high mobility of aerial nodes leads to short channel coherence time, which together with noise causes inevitable channel estimation errors. These errors significantly degrades the performance of wireless channel-based key generation. Therefore, we propose a codeword construction algorithm that achieves a polarization characteristic, which effectively segregates reliable keys from unreliable ones. Compared to the existing quantization-based key generation scheme, our approach maximize the utilization of raw channel information and employ soft-decision decoding to generate key. Simulation results demonstrate that the proposed scheme reduces the key disagreement rate for legitimate users and increases the number of consistently generated keys. Furthermore, our method ensures a lower key consistency rate for eavesdropper, which guarantees system security.
Accurate 3D mapping in endoscopy enables quantitative, holistic lesion characterization within the gastrointestinal (GI) tract, requiring reliable depth and pose estimation. However, endoscopy systems are monocular, and existing methods relying on synthetic datasets or complex models often lack generalizability in challenging endoscopic conditions. We propose a robust self-supervised monocular depth and pose estimation framework that incorporates a Generative Latent Bank and a Variational Autoencoder (VAE). The Generative Latent Bank leverages extensive depth scenes from natural images to condition the depth network, enhancing realism and robustness of depth predictions through latent feature priors. For pose estimation, we reformulate it within a VAE framework, treating pose transitions as latent variables to regularize scale, stabilize z-axis prominence, and improve x-y sensitivity. This dual refinement pipeline enables accurate depth and pose predictions, effectively addressing the GI tract's complex textures and lighting. Extensive evaluations on SimCol and EndoSLAM datasets confirm our framework's superior performance over published self-supervised methods in endoscopic depth and pose estimation.
The FAIR principles have transformed how computational data and workflows are shared in materials research, yet existing repositories can only serve pre-computed entries -- broad coverage is perpetually incomplete and cannot adapt to new questions on demand. To address these challenges, we present OptiMat Alloys, a large language model-powered conversational agent for multi-principal element alloy exploration built on three pillars: a living database that stores every calculation with provenance, low-barrier accessibility through a web interface requiring zero programming expertise, and built-in uncertainty quantification via cross-potential and cross-configuration validation. Coupling foundational machine learning interatomic potentials covering near-all periodic table of elements with natural-language interaction, OptiMat Alloys enables targeted, on-demand computation guided by the user's domain knowledge-extending FAIR from pre-computed repositories to on-demand knowledge generation and making computational alloy screening accessible to any materials scientist.